Dimerization of Terminal Domains in Spiders Silk Proteins Is

Publication Date (Web): April 16, 2015. Copyright ... To understand the terminal domains' dimerization mechanisms, we developed a protein model based ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/journal/abseba

Dimerization of Terminal Domains in Spiders Silk Proteins Is Controlled by Electrostatic Anisotropy and Modulated by Hydrophobic Patches Anıl Kurut,*,† Cedric Dicko,*,‡ and Mikael Lund*,† †

Division of Theoretical Chemistry, Chemical Center, and ‡Division of Pure and Applied Biochemistry, Chemical Center, Lund University, SE-22100 Lund, Sweden S Supporting Information *

ABSTRACT: The well-tuned spinning technology from spiders has attracted many researchers with the promise of producing high-performance, biocompatible, and yet biodegradable fibers. So far, the intricate chemistry and rheology of spinning have eluded us. A breakthrough was achieved recently, when the 3D structures of the N and C terminal domains of spider dragline silk were resolved and their pH-induced dimerization was revealed. To understand the terminal domains’ dimerization mechanisms, we developed a protein model based on the experimental structures that reproduces charge and hydrophobic anisotropy of the complex protein surfaces. Monte Carlo simulations were used to study the thermodynamic dimerization of the N-terminal domain as a function of pH and ionic strength. We show that the hydrophobic and electrostatic anisotropies of the N-terminal domain cooperate constructively in the association process. The dipolar attractions at pH 6 lead to weakly bound dimers by forcing an antiparallel monomer orientation, stabilized by hydrophobic locking at close separations. Elevated salt concentrations reduce the thermodynamic dimerization constant due to screened electrostatic dipolar attraction. Moreover, the mutations on ionizable residues reveal a free energy of binding, proportional to the dipole moment of the mutants. It has previously been shown that dimers, formed at pH 6, completely dissociate at pH 7, which is thought to be due to altered protein charges. In contrast, our study indicates that the pH increase has no influence on the charge distribution of the N-terminal domain. Instead, the pH-induced dissociation is due to an adapted, loose conformation at pH 7, which significantly hampers both electrostatic and hydrophobic attractive interactions. KEYWORDS: silk assembly, electrostatic anisotropy, hydrophobic patchiness, dimer formation, Monte Carlo simulation, silk termini domains



INTRODUCTION Silks have evolved to be some of nature’s most impressive composite materials.1,2 Silk fibers3 are not only one of the toughest polymers known, but they have a number of other characteristics, such as biocompatibility and degradability, that make them an interesting as well as an important object for research in the general areas of biomaterials,4 biomimetic and the coevolution of behavior,5 morphology, and function.2 Harnessing silk production and processing have however been proven difficult and elusive. A complete understanding of silk fiber formation is needed. Spiders have the ability to control key aspects of silk production,1,6 namely (i) silk protein composition, (ii) storage, and (iii) fiber processing. Modern studies on mechanical strength of silks and spinning behavior7−9 show a clear correlation between silk strength and relative elasticity with specific molecular architecture.10,11 Genetic analysis,12−14 and earlier amino acid composition,15 of silk proteins sequences demonstrated a strong bias toward specific amino acid content dominated by alanine and glycine residues. Furthermore, silk gene molecular © XXXX American Chemical Society

design show preferential motif arrangements (polyalanine crystalline domains and glycine-rich amorphous domains) that can be correlated to macroscopic properties of silk such as elasticity and strength.14 Inside the glands, the silk proteins can be stored for long periods of time in an apparently stable form.16,17 This conformation seems to change just prior to fiber formation in order to allow the highly energy efficient conversion from their aqueous solution state to the insoluble fibrillar state.6 For this the individual silk molecules must adopt shapes and conformations dictated by polar and nonpolar moieties.18 To achieve bulk orientations with optimal axial stiffness (which is important for mechanical properties of the fibers), the system must allow the molecules to self-organize in their extended configuration19 followed by an efficient intermolecular lock-in but with only a limited number of cross-links in order to Received: September 30, 2014 Accepted: April 16, 2015

A

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering maintain flow viscosity. However, β-strands (unlike α-helical structures) are unstable in solution as isolated secondary structures20,21 and hence tend to engage and stabilize interactions with neighboring strands, leading to the construction of an intermolecular gelation network.6 To control such “structurally reactive” silk proteins, progressive and judicious modifications of solution conditions (pH, ionic strength, etc.) are necessary.21 The detailed exploration of the chemical controls has highlighted the role of acidification of the silk proteins as a key step prior to fiber formation.21−24 The exact role of this acidification was unclear until recently when the “N-, and C-terminal domains” structures of dragline silk were resolved.25,26 Specifically, the N-terminal domain plays a critical role in delaying aggregation at neutral pH. At pH values of around 6, the N-terminal domain is found as a homodimer, whereas at high salt concentration and neutral pH, it is found as a monomer.27−30 Several models have now been proposed to explain the role of the terminal domains in silk fiber formation29,31,32 via micelles formation and activation of specific amino acids. The details of the interactions involved in the assembly of the N-terminal homodimer and its implication for spinning may not be resolved experimentally. Computational methods can provide new insights on the balance of forces and sequence of events leading to fiber formation. In the present contribution, we model the N and C terminal domains of silk protein based on their available crystallographic and NMR structures. Using these models in molecular computer simulations, we seek to address the following questions: (1) What are the microscopic driving mechanisms for the association of N-terminal domains? (2) What is the effect of electrostatic and hydrophobic interactions? (3) How do solution pH and salt concentration promote or hinder dimerization?



in describing shape and electrostatic features, whereas short-ranged attractions, such as van der Waals and hydrophobic interactions, require simple assumptions about their strengths. In this work, we use generic values for these strengths and have made no adjustments for the current protein system. Although a detailed model description is provided in the references above, a brief introduction is given: The amino acid beads are placed at their mass centers as obtained from PDB 3LR225 and PDB 2KHM,26 for the N and C terminal domains, respectively. Each bead or particle can be of three distinct types: (i) charged, (ii) neutral, or (iii) hydrophobic, which controls the intermolecular interactions via the following system energy function excluded volume and vdW energy

βUtot

electrostatic energy

  12 ⎛⎛  ⎛ σij ⎞6 ⎞  σij ⎞ λBziz j −κrij ⎜ = ∑ 4β ϵLJ⎜⎜⎜ ⎟⎟ − ⎜⎜ ⎟⎟ ⎟⎟ + ∑ e r ⎝ rij ⎠ ⎠ i > j ∈ nc rij i,j∈n ⎝⎝ ij ⎠ intrinsic protonation energy

hydrophobic energy

   + ∑ −  βγ ΔAsasa (pK a,i − pH)ln 10 eff i ∈ np

(1)

where the indices n, nc, and np run over all, charged, and protonated particles, respectively. β = 1/kBT is the thermal energy; σij = (σi + σj)/2 is the Lennard−Jones diameter for a pair of residues, i and j; βϵLJ = 0.05 is an empirically determined interaction strength describing short-ranged attraction;33,34,37 λB = 7 Å the Bjerrum length for water at 298 K; z residue valencies; rij the interparticle separation; κ = (8πλB∑ionsckz2k) 1/2 is the inverse Debye length; ck the number density of mobile ion type k; pKa,i are intrinsic acid dissociation constants for acidic and basic residues c.f. Table 2; pH is the solution acidity. Finally, γeff is an effective surface tension controlling the attraction between hydrophobic groups based on the change in solvent accessible surface area, Asasa, upon interparticle approach. Configurations in the statistical thermodynamic NVT ensemble are sampled using Metropolis Monte Carlo simulations of one or two proteins where we average over orientations, protein−protein separations and protonation states (see Figure 1) with a move frequency 1:1:nc. A Boltzmann distribution of protonation states is sampled via particle swap moves between protonated and deprotonated forms of acidic and basic residues.35 All thermal averaging are taken over 45 × 106 configurations, and preceded by 1 × 105 equilibration configurations. In the case of two proteins, these are placed on a line connecting their mass centers and during simulation we sample the pair distribution function, g(R), as well as electric multipole interactions (see Table 1) as a function of mass center separation, R. The angularly averaged potential of mean force, βw(R) = −ln g(R) is used to estimate relative protein−protein binding free energies between the wild-type and mutated forms

MODEL AND METHODOLOGY

To model the globular N and C terminal domains of spiders silk, we use a coarse graining procedure where the protein structures are described as rigid bodies build from beads representing amino acid residues, see Figure 1. This structural model has previously been used

β ΔΔG = − ln

mut Kdimer wild Kdimer

, ∞

Kdimer = 4π Figure 1. Illustration of spider silk where the C and N terminal domains are connected by a peptide sequence (purple, not modeled). The proteins are built from spheres that can be neutral hydrophobic (gray), or charged (blue or red). During simulation, the proteins translate (red arrow) and rotate (green arrows) and protons fluctuate (blue arrow). Water and dissolved salt modulate interparticle interactions through the dielectric constant, ϵr, and the Debye screening length, 1/κ.

∫9

(exp(− βw(R )) − 1)R2dR

(2)

where Kdimer is the thermodynamic association constant38 and 9 is the distance of closest approach between two monomers. Note that the Metropolis Monte Carlo method reports only on equilibrium properties, not kinetics. Solvent-accessible surface areas, Asasa, for hydrophobic residues (ALA, ILE, LEU, MET, PHE, PRO, TRP, VAL) were measured in the atomistic experimental structures by rolling a spherical probe of radius 1.5 Å over the surface.40 The surfaces of two residues were considered inaccessible when they were closer than 3 Å, resulting in ΔAsasa = ∑i ∈ ninaccAi. The effective surface tension is set to 2.97 dyn/cm in 10 mM NaCl solution, corresponding to a 0.44 kBT attraction for a residue pair with an averaged sized Asasa of 30 Å2 (see Table 3).

to interpret experimentally observed phase diagrams, osmotic second virial coefficients, and solution structure factors for several protein systems.33−36 Importantly, the model is free for adjustable parameters B

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Table 1. Electric Multipole Expansion up to Third Order for Two Charge Distributions, A and B, Connected along the z-Axis37a component

energy expressions

multipole−multipole ion−ion ion−dipole ion−quadrupole dipole−dipole

∑i A∑j B ((qiqj)/(rij)) N N ((∑i Aqi∑j Bqj)/(RAB)) N N N N −((∑i Aqi∑j Bqjzj)/(R2AB))+((∑i Aqizi∑j Bqj)/(R2AB)) NA NB N N 2 2 2 3 ((∑i qi∑j qj(zj −1/2yj −1/2xj ))/(RAB))+((∑j Bqj∑i Aqi(z2i −1/2y2i −1/2x2i ))/(R3AB)) NA NB NA NB N 3 3 −2((∑i qizi∑j qjzj)/(RAB))+((∑i qiyi∑j qjyj)/(RAB))+((∑i Aqixi∑Nj Bqjxj)/(R3AB))

N

N

a

Note that the equations are valid when the protein−protein center of mass separation, RAB is larger than the protein radius, RAB > Rmax = 23 Å in the limit of no or low salt. x, y, and z refer to x−, y− and z− component of amino acid positions with respect to protein centre of mass.

Elevated salt concentrations allow higher charges on the clustered residues due to screened repulsion between them (Table 4). Therefore, the proteins can achieve the same level of deprotonation at a slightly lower pH value. This influence is magnified in the dipole moment due to the collective effect of the clustered charges. Note that the charge of LYS and ARG in the clusters are already fully protonated even at low salt concentrations, thus their protonation states are unaffected by salt. In contrast, the C-terminal domain has an isotropic distribution of a few charges, resulting in a weakened dipole moment, insensitive to salt. Hydrophobic Anisotropy. The hydrophobic anisotropy of a protein is caused by the clustering of solvent exposed hydrophobic residues on the surface. There is no obvious equivalent to the multipole expansion performed for charges and we instead devised a scheme where we located the three largest hydrophobic patches on the two terminal domains using solvent accessible surface areas, Asasa. By this analysis, both domains are found to have large hydrophobic, exposed surfaces that are reduced upon dimerization−see Table 5 and Figure 2. In the C-terminal domain, 40% of the exposed surface is hydrophobic which is roughly twice that of similarly sized globular proteins such as lysozyme and α-lactalbumin. This indicates a high preference for hydrophobic association. As seen in Figure 3b, the C-terminal domain has three large patches in which the largest (h1) encompass more than 40% of the exposed hydrophobic surface. The N-terminal domain has a slightly lower hydrophobic anisotropy and the largest patch constitutes 29% of the hydrophobic surface. Anisotropy in Dimers. Next, we investigated how the dipole moment vector and the hydrophobic patch vectors are oriented in the experimental dimer structures, see Figure 4. To quantify the monomer orientations in the dimers, we calculated the scalar products of the patch and dipole vectors, see Figure 5. These scalar products were later used as a reference to compare simulated dimer structures with the experimental structures. The monomer orientation in the N-terminal domain dimers ensures an antiparallel dipole alignment, yielding a scalar product of dipole moments, (d−d ≈ −1). However, this orientation suppresses ion−dipole interactions, (i−d ≈ 0). Because ASP, LYS, and ARG residues are situated on the binding surface, the antiparallel monomer orientation enables close contact salt bridges between the ASP residues of one monomer with LYS and ARG residues of the other monomer in the dimeric form. The monomer orientation in the NMR structure of the C-terminal domain dimer also establishes antiparallel dipole moments at pH 6, (d−d ≈ −1). However, as mentioned in the previous section the C-terminal domain dipole is insufficient to achieve this alignment, implying a cooperation with the hydrophobic anisotropy. This configuration, indeed, maximizes the hydrophobic interactions by providing a full correlation

Table 2. Intrinsic Acid Dissociation Constants (pKa) of Ionizable (Titratable) Residues Including Carboxyl (CTR) and Amino (NTR) Groups of the Terminal Residues of Each Domain39 number site

pKa

N-ter

C-ter

ASP GLU HIS TYR LYS CYS ARG CTR NTR nc

4.0 4.4 6.3 9.6 10.4 10.8 12 2.6 7.5

2 5 1 0 2 0 1 1 1 13

1 1 0 2 0 1 2 1 1 9

Table 3. Effective Surface Tension and the Change in Macroscopic Surface Tension of Water41 (∂γ/∂csalt)

a

salt

∂γ/∂csalt (mN/m)/(mol/L)

γeff (10 mM) (dyn/cm)a

γeff (100 mM) (dyn/cm)a

NaCl

1.70

2.97

3.12

γeff = γeff(0 M) + csalt∂γ/∂csalt.



RESULTS AND DISCUSSIONS Structural Analysis of the Terminal Domains. We first investigated how the charged and hydrophobic residues are distributed on the surface of the N- and C-terminal domains of the silk protein to determine surface anisotropy or “patchiness”. Later, we quantified how these patches are located on the experimental structures of the homodimers. Electrostatic Anisotropy. Electrostatic anisotropy is brought about by an uneven distribution of charge on the protein surface. Such “patchiness” can be mathematically described using multipoles (see Table 1) and an increasing number of termsmonopole, dipole, quadrupole, ...increases the precision. We analyzed the monopole and dipole moments of the N and C terminal domains as a function of salt concentration and pH to give a qualitative picture of the leading electrostatic terms. As seen in Figure 2, the N-terminal domain charge decreases rapidly from +4.5e to −4e when the solution pH is elevated from 2 to 6. This is accompanied by a dramatic increase of the dipole moment, peaking at pH 6. The charge distribution of the N-terminal domain shows a remarkable anisotropy formed by separated clusters of GLU/ASP and HIS/ARG/LYS, respectively. This results in a significant dipole moment with a well-defined direction (Figure 2c). These clustered residues yield a maximized dipole moment around pH 6 where dimerization is favored.27,42 C

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Figure 2. (a) Net charge and (b) dipole moment of terminal domains as a function of pH and 1:1 salt concentration; (c, d) show the dimerization surface with hydrophobic residues depicted in gray and charged residues at pH 6 are shown in red and blue. Arrows indicate dipole moment directions of the N and C terminal domains and the arrow thicknesses are proportional to the dipolar strengths.

Table 4. Effect of Salt Screening on the Average Charge of Clustered Ionizable Residues at pH 6 positive end

negative end

residue

10 mM

100 mM

HIS6 LYS65 ARG60 ASP39 ASP40 ASP134 GLU79 GLU84 GLU85 GLU119

0.151 1.000 1.000 −0.921 −0.876 −0.778 −0.858 −0.770 −0.887 −0.966

0.173 1.000 1.000 −0.960 −0.943 −0.886 −0.932 −0.896 −0.943 −0.973

Figure 3. Three largest hydrophobic patches of the N and C terminal domains of silk protein. The arrows point to the surface centers of the patches.

tweezer-like interactions between the helix 1 of chain A and helix 5 of chain B.26 Because the tweezer-like entanglement is inaccessible without unfolding the monomer, our rigid protein model is inadequate to study C-terminal domain dimerization. In the following, we therefore focus merely on the N-terminal domain. N-Terminal Domain Dimerization in Salt Solutions. We next studied the interplay between electrostatic and hydrophobic interactions in the dimerization mechanism. This was done first by simulating the association with and without hydrophobic interactions. Then, we investigated the effect of pH and salt concentration. We also examined the relation between the dipole moment and the association constant via a mutation study on ionizable groups of the N-terminal domain. Effect of Hydrophobic Interactions. Experiments show that the N-terminal domain self-assembles into a stable dimer at pH 6 and at low salt concentrations which is diminished by elevated pH and salt,27,42 hinting at the importance of electrostatics. We first performed dimerization simulations at pH 6 in 10 mM salt solution by considering only electrostatic and vdW interactions. As shown in Figure 5, the affinity

Table 5. Surface Properties of the Crystal and NMR Structures of the N and C Terminal Domainsa N-terminal D. residue number, n number of hydrophobic residues total surface area, Atot, Å2 hydrophobic surface in % of Atot binding surface per monomer, Å2 hydrophobic % of the binding surface

C-terminal D.

monomer

dimer

monomer

dimer

125 57 6192 28% 1475 41%

250 114 9812 24%

108 51 6512 38% 2548 55%

216 102 7918 29%

a

Solvent-accessible surface areas were calculated with a 1.5 Å probe, including both side chain and backbone of the solvent exposed residues.

between the first and the third largest patches (h11 ≈ h33 ≈ −1), resulting in exclusion of these patches from the solvent upon dimerization. It has been shown that C-terminal domain dimerization involves a disulfide bond between the monomers and D

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Table 6. Difference in Calculated Binding Free Energies of the N-Terminal Domain in kcal/mol at Various Salt Concentrations and pH, with respect to the Affinity at pH 6 and 10 mM, See eq 2 salt conc. 10 mM 100 mM

pH 2a

pH 6

pH 8

0 0.69

1.26 1.16

a

At pH 2, there is a net-repulsion between the proteins and the binding constant is thus ill-defined.

pH 6 and is reduced by elevated pH and salt concentrations as shown in Table 6. Moreover, Figure 6 shows that dimerization is diminished at pH 2 because of the protonation of negative ASP and GLU residues. The interaction free energy has two minima at low salt concentrations, suggesting two modes of dimerization. The minimum at 23 Å separation is due to strong electrostatic attraction, whereas at 18 Å hydrophobic and vdW attractions are maximized. However, at high salt concentration, hydrophobic interactions dominate over the screened electrostatics. Note that even for the minimum at 23 Å, electrostatic and hydrophobic interactions cooperate and contribute similarly to the binding free energy. As seen in Figure 6, third row, increasing pH from 6 to 8, affects mainly dipole−dipole interactions and together with the enhanced net charge repulsion, this leads to a reduced binding affinity. As shown in Figure 2, the dipole moments and net charges at pH 6 and pH 7 are similar, suggesting that this pH change will have only a small effect on the dimerization of the N-terminal domain. Effect of Mutations. To quantify the dipole moment effect on the binding affinity, we simulated the dimerization of the following mutated forms of the N terminal domain: • single mutations: D40N, E79Q, and E84Q. • double mutation: D40NE84Q. • deletion mutation: H6-. All mutations, except deletion of H6, reduce the dipole moment of the N-terminal domain at pH 6 while preserving its direction (Figure 7). The most dramatic reduction of 25% is caused by the double mutation D40NE84Q because of the simultaneous elimination of the two charges from the negative end. Calculated dipole moments of the mutants at pH 6 follows

Figure 4. Hydrophobic and dipole moment alignments in the crystal structure of dimers. Chains A and B are shown in blue and pink, respectively.

without hydrophobic interactions is insufficient to form strongly bound dimers. However, with an additional shortrange attraction of hydrophobic origin (γ), the N-terminal domain forms well-defined dimers where the monomer orientation exactly matches the crystal structure, see Figure 5b, c. This indicates that the hydrophobic and electrostatic interactions cooperate constructively to form stable dimers. Further confirmation of this mechanism is found in experiments where mutating a centrally located hydrophobic residue on the binding surface to a negatively charged ARG (A72R), diminishes dimer formation.30 Effect of Salt Concentration and pH. In line with experiments,27,42 self-association of the N-terminal domain occurs at

Figure 5. (a) Angularly averaged interaction free energy of the N-terminal domain with and without hydrophobic attraction. (b) Comparison between crystal and simulated alignment of hydrophobic and dipole vectors at pH 6, 10 mM salt with (black) and without (red) hydrophobic attraction. (c) Preferred locations of HIS6 (pink bubble), ASP40 (purple bubble), GLU79 (green bubble), and GLU84 (blue bubble) of chain B in the simulated dimers and in the crystal structure (shaded and dotted spheres with corresponding colors). Chain A in the crystal and in the simulated dimers are located on top of each other and depicted by a cartoon and solid ball representations, respectively. E

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Figure 6. Association of N-terminal domains as a function of protein center of mass separation at 10 mM (solid lines) and 100 mM (dotted lines) ionic strength. Note that the multipole interaction energies are only represented at the low salt condition and when the separation is larger than radius of the protein where the mutipole expansions are valid.

Figure 7. Electrostatic properties of mutated N-terminal domains as a function of pH at 10 mM salt concentration. Each mutation is color-marked identically on the legends and on the structure.

Table 7. Difference in Binding Free Energy of Mutated N-Terminal Domains with Respect to the Wild-Type at Corresponding pH and 10 mM Ionic Strength (ΔΔG = ΔGmut − ΔGwt) in Units of kcal/mol experimenta,29 simulation

pH

H6-

D40N

E79Q

E84Q

double

wild

6 6 8

1.0 ± 0.2b 0.35 −0.04

n.d.c 0.35 0.51

0.3 ± 0.2 0.19 0.05

0.6 ± 0.1 0.21 0.24

n.a.d 0.85 0.65

0 0 0

a Experiments are performed at 50 mM ionic strength. bHIS6 replaced with ALA residue instead of deletion. cn.a. = not available. dn.d. = no measurable dimerization.

the order μwt ≈ μH6 > μE79 = μD40N > μE84 > μD40NE84Q. This correlates very well with the calculated binding constants,

E79 E84 D40N H6 D40NE84Q Kwt , suggestdimer > Kdimer >Kdimer > Kdimer = Kdimer > Kdimer ing a direct connection between the dipole moment and the

F

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering

Figure 8. ΔΔG as a function of dipole strength of mutated N-terminals at 10 mM ionic strength.

binding affinity (Table 7 and Figure 8). This predicted order closely follows that observed experimentally.29,42 The double mutation D40NE84Q has the weakest dipole moment and causes the most dramatic reduction in the electrostatic attraction (see Supporting Information), resulting in a significantly suppressed binding constant. Note that the mutation of ASP (D40N) has a more pronounced effect on the binding affinity than the mutation of GLU79, due to the broken salt bridge attractions between ASP40 and ARG60/LYS65. However, the Debye−Hückel approximation tends to overestimate salt screening between close contact charges, leading to less pronounced ΔΔG reductions, especially when mutating the salt bridge forming residues like ASP40. Although the dipole moment seems insensitive to the H6- mutation when pH is higher than 5, an affinity reduction similar to the D40N mutation is observed at pH 6 but not at pH 8 (Table 7). This is because H6 has the ability to charge itself up as a response to negative charges in close vicinity when solution pH approaches pKa,HIS ≈ 6. Hence, HIS contributes to the dipole moment and provides an additional attraction only at pH 6 and in the vicinity of negative charges, but neither at pH 8 nor in the bulk alone without any nearby negative charges. It should be noted that our rigid protein model neglects possible conformational changes in the wild type caused by the mutations and hence only approximates electrostatically induced effects. pH-Induced Conformational Change. It has recently been shown that the N-terminal domain adopts two different structures depending on the environmental pH.43 At pH 6, it obtains a compact conformation, identical to the crystal structure. At pH 7, however, a loose conformation (PDB 2LPJ) is obtained by pushing a helix bundle on the dimerization site outward. This distorts the binding surface such that the helix bundle separates some of the hydrophobic residues from the charged ones; see ref 43 for a detailed comparison of the two structures. As seen in Figure 9, at pH 6 and 7, this conformational change has a negligible influence on the charge distribution and both the dipole moment strength and direction are unaffected. To determine the effect of the structural change on the binding affinity of the N-terminal domain, we investigated the dimerization of the loose conformation and compared it with the compact one. The association of the loose conformation is studied at pH 6 and low salt concentrations to ensure favorable electrostatic conditions for a possible dimerization. Figure 9a shows that the monomer−monomer attraction is significantly reduced when the loose conformation is adopted, resulting in a ΔΔG = 1.68 kcal/mol lower binding

Figure 9. (a) Dimerization free energy of compact and loose conformation of N-terminal domain at pH 6 as a function of protein center of mass separation and their dipole moments (inset) as a function of pH. (b) Interaction energy components during dimerization. (c, d) Compact and loose protein conformations, respectively, with color-coded hydrophobic (gray) and charged (red and blue) surfaces. The arrows depict the direction of the dipole moment at various pH.

free energy. This suppression originates from the reduction of both hydrophobic and electrostatic energies, implying a destructive cooperation of the hydrophobic and electrostatic anisotropies.



CONCLUSION We have presented a dimerization study of terminal domains of silk proteins to understand the interplay between electrostatic and hydrophobic interactions as a function of pH and salt concentration. In the dimerization process of the N-terminal domain, electrostatic and hydrophobic interactions cooperate such that the dimerization proceeds via electrostatic dipole interactions, forcing an antiparallel monomer orientation, followed by hydrophobic interaction-induced locking of the dimer. In agreement with experimental observations, the strongest monomer binding is predicted at pH 6 and low salt concentrations. Elevated salt concentrations and pH suppress the electrostatic dipole attraction and weakens binding. Point mutations of ionizable groups show that the association constant of N-terminal domain is proportional to the dipole moment. Our results further indicate that the hindered dimerization at pH 7 is not because of the altered charge distribution, but because of the adopted conformation. The conformation at pH 7 destroys the constructive cooperation between electrostatic and hydrophobic interactions and significantly reduces the binding. Interestingly, the analysis on the surface properties of the C-terminal domain monomer has indicated a strong influence of hydrophobic interactions on the assembly rather than electrostatics. The large hydrophobic surface of the C-terminal domain and the possible disulfide bond between the monomers support the idea that the C-terminal domain is dimerized during protein secretion in the spider gland. G

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering



(17) Inoue, S.; Tanaka, K.; Arisaka, F.; Kimura, S.; Ohtomo, K.; Mizuno, S. Silk fibroin of Bombyx mori is secreted, assembling a high molecular mass elementary unit consisting of H-chain, L-chain, and P25, with a 6:6:1 molar ratio. J. Biol. Chem. 2000, 275, 40517−28. (18) Jin, H. J.; Kaplan, D. L. Mechanism of silk processing in insects and spiders. Nature (London, U. K.) 2003, 424, 1057−61. (19) Donald, A. M.; Windle, A. H.; Hanna, S. Liquid Crystalline Polymers, 2nd ed.; Cambridge University Press: Cambridge, U.K., 1992. (20) Nesloney, C. L.; Kelly, J. W. Progress Towards Understanding [beta-Sheet Structure. Bioorg. Med. Chem. 1996, 4, 739−766. (21) Dicko, C.; Kenney, J. M.; Vollrath, F. Beta-silks: enhancing and controlling aggregation. Adv. Protein Chem. 2006, 73, 17−53. (22) Vollrath, F.; Knight, D. P.; Hu, X. W. Silk production in a spider involves acid bath treatment. Proc. R. Soc. London, Ser. B 1998, 265, 817−820. (23) Dicko, C.; Kenney, J. M.; Knight, D.; Vollrath, F. Transition to a beta-sheet-rich structure in spidroin in vitro: the effects of pH and cations. Biochemistry 2004, 43, 14080−7. (24) Dicko, C.; Vollrath, F.; Kenney, J. M. Spider silk protein refolding is controlled by changing pH. Biomacromolecules 2004, 5, 704−10. (25) Askarieh, G.; Hedhammar, M.; Nordling, K.; Saenz, A.; Casals, C.; Rising, A.; Johansson, J.; Knight, S. D. Self-assembly of spider silk proteins is controlled by a pH-sensitive relay. Nature (London, U. K.) 2010, 465, 236−8. (26) Hagn, F.; Eisoldt, L.; Hardy, J. G.; Vendrely, C.; Coles, M.; Scheibel, T.; Kessler, H. A conserved spider silk domain acts as a molecular switch that controls fibre assembly. Nature (London, U. K.) 2010, 465, 239−42. (27) Hagn, F.; Thamm, C.; Scheibel, T.; Kessler, H. pH-dependent dimerization and salt-dependent stabilization of the N-terminal domain of spider dragline silk-implications for fiber formation. Angew. Chem., Int. Ed. 2011, 50, 310−3. (28) Wallace, J.; Shen, J. Unraveling A Trap-and-Trigger Mechanism in the pH-Sensitive Self-Assembly of Spider Silk Proteins. J. Phys. Chem. Lett. 2012, 3, 658−662. (29) Schwarze, S.; Zwettler, F. U.; Johnson, C. M.; Neuweiler, H. The N-terminal domains of spider silk proteins assemble ultrafast and protected from charge screening. Nat. Commun. 2013, 4, 2815. (30) Kronqvist, N.; Otikovs, M.; Chmyrov, V.; Chen, G.; Andersson, M.; Nordling, K.; Landreh, M.; Sarr, M.; Jörnvall, H.; Wennmalm, S.; Widengren, J.; Meng, Q.; Rising, A.; Otzen, D.; Knight, S. D.; Jaudzems, K.; Johnasson, J. Sequential pH-driven dimerization and stabilization of the N-terminal domain enables rapid spider silk formation. Nat. Commun. 2014, 5, 3254. (31) Silvers, R.; Buhr, F.; Schwalbe, H. The molecular mechanism of spider-silk formation. Angew. Chem., Int. Ed. 2010, 49, 5410−2. (32) Eisoldt, L.; Thamm, C.; Scheibel, T. Review the role of terminal domains during storage and assembly of spider silk proteins. Biopolymers 2012, 97, 355−61. (33) Salvatore, D. B.; Duraffourg, N.; Favier, A.; Persson, B. A.; Lund, M.; Delage, M.-M.; Silvers, R.; Schwalbe, H.; Croguennec, T.; Bouhallab, S.; Forge, V. Investigation at residue level of the early steps during the assembly of two proteins into supramolecular objects. Biomacromolecules 2011, 12, 2200−10. (34) Kurut, A.; Persson, B. A.; Åkesson, T.; Forsman, J.; Lund, M. Anisotropic Interactions in Protein Mixtures: Self Assembly and Phase Behavior in Aqueous Solution. J. Phys. Chem. Lett. 2012, 3, 731−734. (35) Lund, M.; Jönsson, B. Charge regulation in biomolecular solution. Q. Rev. Biophys. 2013, 46, 265−81. (36) Kaieda, S.; Lund, M.; Plivelic, T. S.; Halle, B. Weak SelfInteractions of Globular Proteins Studied by Small-Angle X-ray Scattering and Structure-Based Modeling. J. Phys. Chem. B 2014, 118, 10111−9. (37) Persson, B. A.; Jönsson, B.; Lund, M. Enhanced protein steering: cooperative electrostatic and van der Waals forces in antigen-antibody complexes. J. Phys. Chem. B 2009, 113, 10459−64.

ASSOCIATED CONTENT

S Supporting Information *

The following file is available free of charge on the ACS Publications website at DOI: 10.1021/ab500039q. Definition of hydrophobic patches and their corresponding vectors as well as a detailed energy analysis of the mutant associations (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS For support and resources the authors thank the eSSENCE strategic program; the LUNARC center for scientific computing; the Linneaus Center “Organizing Molecular Matter”; the Swedish Research Council; and the Swedish Strategic Research Foundation.



REFERENCES

(1) Knight, D.; Vollrath, F. Changes in element composition along the spinning duct in a Nephila spider. Naturwissenschaften 2001, 88, 179−182. (2) Craig, C. L. Spiderwebs and Silks: Tracing Evolution from Molecules to Genes to Phenotypes; Oxford University Press: New York, 2003. (3) Porter, D.; Guan, J.; Vollrath, F. Spider silk: super material or thin fibre? Adv. Mater. (Weinheim, Ger.) 2013, 25, 1275−9. (4) Altman, G. H.; Diaz, F.; Jakuba, C.; Calabro, T.; Horan, R. L.; Chen, J.; Lu, H.; Richmond, J.; Kaplan, D. L. Silk-based biomaterials. Biomaterials 2003, 24, 401−16. (5) Vollrath, F.; Selden, P. The Role of Behavior in the Evolution of Spiders, Silks, and Webs. Annu. Rev. Ecol. Evol. Syst. 2007, 38, 819− 846. (6) Vollrath, F.; Porter, D.; Holland, C. The science of silks. MRS Bull. 2013, 38, 73−80. (7) Vollrath, F. Strength and structure of spiders’ silks. Rev. Mol. Biotechnol. 2000, 74, 67−83. (8) Vollrath, F.; Porter, D. Spider silk as archetypal protein elastomer. Soft Matter 2006, 2, 377. (9) Vollrath, F.; Porter, D.; Holland, C. There are many more lessons still to be learned from spider silks. Soft Matter 2011, 7, 9595. (10) Riekel, C.; Madsen, B.; Knight, D.; Vollrath, F. X-ray diffraction on spider silk during controlled extrusion under a synchrotron radiation X-ray beam. Biomacromolecules 2000, 1, 622−6. (11) van Beek, J. D.; Hess, S.; Vollrath, F.; Meier, B. H. The molecular structure of spider dragline silk: folding and orientation of the protein backbone. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 10266−71. (12) Xu, M.; Lewis, R. V. Structure of a protein superfiber: spider dragline silk. Proc. Natl. Acad. Sci. U. S. A. 1990, 87, 7120−4. (13) Guerette, P. A.; Ginzinger, D. G.; Weber, B. H.; Gosline, J. M. Silk properties determined by gland-specific expression of a spider fibroin gene family. Science (Washington, DC, U. S.) 1996, 272, 112− 115. (14) Hayashi, C. Y.; Shipley, N. H.; Lewis, R. V. Hypotheses that correlate the sequence, structure, and mechanical properties of spider silk proteins. Int. J. Biol. Macromol. 1999, 24, 271−5. (15) Andersen, S. O. Amino acid composition of spider silk. Comparative Biochemistry and Physiology 1970, 35, 705−711. (16) Knight, D.; Vollrath, F. Hexagonal columnar liquid crystal in the cells secreting spider silk. Tissue Cell 1999, 31, 617−20. H

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX

Article

ACS Biomaterials Science & Engineering (38) Evans, D. F.; Wennerström, H. The Colloidal Domain: Where Physics, Chemistry, Biology, and Technology Meet, 2nd ed.;Wiley−VCH: New York, 1999. (39) Nozaki, Y., Tanford, C. In Enzyme Structure; Hirs, C. H. W., Ed.; Methods in Enzymology; Academic Press: New York, 1967; Vol. 11, pp 715−734. (40) Humphrey, W.; Dalke, a.; Schulten, K. VMD: visual molecular dynamics. J. Mol. Graphics 1996, 14 (1), 33−38. (41) Boström, M.; Williams, D. R. M.; Ninham, B. W. Surface Tension of Electrolytes: Specific Ion Effects Explained by Dispersion Forces. Langmuir 2001, 17, 4475−4478. (42) Landreh, M.; Askarieh, G.; Nordling, K.; Hedhammar, M.; Rising, A.; Casals, C.; Astorga-Wells, J.; Alvelius, G.; Knight, S. D.; Johansson, J.; Jörnvall, H.; Bergman, T. A pH-dependent dimer lock in spider silk protein. J. Mol. Biol. 2010, 404, 328−36. (43) Jaudzems, K.; Askarieh, G.; Landreh, M.; Nordling, K.; Hedhammar, M.; Jörnvall, H.; Rising, A.; Knight, S. D.; Johansson, J. pH-dependent dimerization of spider silk N-terminal domain requires relocation of a wedged tryptophan side chain. J. Mol. Biol. 2012, 422, 477−87.

I

DOI: 10.1021/ab500039q ACS Biomater. Sci. Eng. XXXX, XXX, XXX−XXX